svclassify: a method to establish benchmark structural variant calls

Parikh, Hemang; Iyer, Hariharan; Chen, Desu; Pratt, M. R.; Bartha, Gábor; Spies, Noah; Losert, Wolfgang; Zook, Justin M.; Salit, Marc

doi:10.1101/019372

Cited by 47 publications

(90 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…27,28 A subset of CNVs from NA12878 were confirmed and further refined to those with support from multiple technologies using SVClassify . 29 The unique collection of Sanger sequencing from the HuRef sample has also been used to characterize SVs. 30,31 Long reads were used to broadly characterize SVs in a haploid hydatidiform mole cell line.…”

Section: Introductionmentioning

confidence: 99%

A robust benchmark for germline structural variant detection

Zook¹,

Nf²,

Nd³

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

New technologies and analysis methods are enabling genomic structural variants (SVs) to be detected with ever-increasing accuracy, resolution, and comprehensiveness. Translating these methods to routine research and clinical practice requires robust benchmark sets. We developed the first benchmark set for identification of both false negative and false positive germline SVs, which complements recent efforts emphasizing increasingly comprehensive characterization of SVs. To create this benchmark for a broadly consented son in a Personal Genome Project trio with broadly available cells and DNA, the Genome in a Bottle (GIAB) Consortium integrated 19 sequence-resolved variant calling methods, both alignment-and de novo assembly-based, from short-, linked-, and long-read sequencing, as well as optical and electronic mapping. The final benchmark set contains 12745 isolated, sequence-resolved insertion and deletion calls ≥50 base pairs (bp) discovered by at least 2 technologies or 5 callsets, genotyped as heterozygous or homozygous variants by long reads. The Tier 1 benchmark regions, for which any extra calls are putative false positives, cover 2.66 Gbp and 9641 SVs supported by at least one diploid assembly. Support for SVs was assessed using svviz with short-, linked-, and long-read sequence data. In general, there was strong support from multiple technologies for the benchmark SVs, with 90 % of the Tier 1 SVs having support in reads from more than one technology. The Mendelian genotype error rate was 0.3 %, and genotype concordance with manual curation was >98.7 %. We demonstrate the utility of the benchmark set by showing it reliably identifies both false negatives and false positives in high-quality SV callsets from short-, linked-, and long-read sequencing and optical mapping. GIAB is working towards a new version of the benchmark set that will use new technologies and methods such as PacBio Circular Consensus Sequencing and ultralong Oxford Nanopore sequencing to expand to more challenging genome regions and include more challenging SVs such as inversions. We are also developing a robust integration process to make calls on GRCh37 and GRCh38 for all seven GIAB samples.

show abstract

Section: Introductionmentioning

confidence: 99%

A robust benchmark for germline structural variant detection

Zook¹,

Nf²,

Nd³

et al. 2019

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…For testing the structural variant callers, we downloaded the Genome in a Bottle high-confidence deletion dataset to use as our truth set against sample NA12878 [17]. At the time of initial pipeline development, Manta v1.0.1 [10] was found to have the highest sensitivity (91%) amongst the structural variant software we internally tested.…”

Section: Structural Variant Callingmentioning

confidence: 99%

Identification of Pathogenic Structural Variants in Rare Disease Patients through Genome Sequencing

et al. 2019

Preprint

View full text Add to dashboard Cite

Purpose: Clinical whole genome sequencing is becoming more common for determining the molecular diagnosis of rare disease. However, standard clinical practice often focuses on small variants such as single nucleotide variants and small insertions/deletions. This leaves a wide range of larger "structural variants" that are not commonly analyzed in patients. Methods: We developed a pipeline for processing structural variants for patients who received whole genome sequencing through the Undiagnosed Diseases Network (UDN). This pipeline called structural variants, stored them in an internal database, and filtered the variants based on internal frequencies and external annotations. The remaining variants were manually inspected and then interesting findings were reported as research variants to clinical sites in the UDN. Results: Of 477 analyzed UDN cases, 286 cases (≈ 60%) received at least one structural variant as a research finding. The variants in 16 cases (≈ 4%) are considered "Certain" or "Highly likely" molecularly diagnosed and another 4 cases are currently in review. Of those 20 cases, at least 13 were identified originally through our pipeline with one finding leading to identification of a new disease. As part of this paper, we have also released the collection of variant calls identified in our cohort along with heterozygous and homozygous call counts. This data is available at https://github.com/HudsonAlpha/UDN_SV_export. Conclusion: Structural variants are key genetic features that should be analyzed during routine clinical genomic analysis. For our UDN patients, structural variants helped solve ≈ 4% of the total number of cases (≈ 13% of all genome sequencing solves), a success rate we expect to improve with better tools and greater understanding of the human genome.

show abstract

“…Up to the point of this work, there was few work focusing on predicting SV segments using base-pair RD information for a single sample. To compare model performances with the SVM model used in the SVclassify work [17], we generated SV-containing and non-SV-containing labels based on the segmentation results for a simplified classification task. The classification performance was evaluated using AUC, sensitivity and false discovery rate (FDR).…”

Section: Evaluation Metricmentioning

confidence: 99%

Enhancing breakpoint resolution with deep segmentation model: a general refinement method for read-depth based structural variant callers

Zhang

Imoto

Miyano

et al. 2019

Preprint

View full text Add to dashboard Cite

Motivation: Structure variants (SVs) are complex genomic alterations with affected genome sizes larger than 50 nucleotides. They have been found in both healthy individuals and patients, which contribute to phenotypic traits and are highly correlated with diseases, such as cancer. With the advance of sequencing technologies, more precise profiling of SVs becomes available. It is still a challenging task to determine SVs in nucleotide-resolution with satisfying both high accuracy and coverage for a single sample. Result: In this paper, we proposed a novel nucleotide-resolution SV detection method for determining SV fragment inside of a bin with base-pair read-depth (RD) information. Distinguished from traditional RD based SV detectors, which usually require grouping and smoothing RD signals at the scale of a bin, we use a U-net model to directly process base-pair RD signals inside of a bin and learn feature representations for intra-bin SV segmentation. The proposed method can be used solely for searching target SVs with RD signature or integrated with previous bin-based SV detection methods to exceed the bin-size limitation. We did a systematic evaluation of the method using WGS data from 1000 Genomes Project. In both single-sample and cross-sample experiments, the U-net model achieves significantly better performances on detecting intra-bin SV fragments when compared with convolutional neural networks (CNN). Furthermore, the model shows a good ability to detect specific SV subtypes, such as ALU deletion.

show abstract

svclassify: a method to establish benchmark structural variant calls

Cited by 47 publications

References 29 publications

A robust benchmark for germline structural variant detection

A robust benchmark for germline structural variant detection

Identification of Pathogenic Structural Variants in Rare Disease Patients through Genome Sequencing

Enhancing breakpoint resolution with deep segmentation model: a general refinement method for read-depth based structural variant callers

Contact Info

Product

Resources

About